Research on Supply Chain Optimization Based on Lstm Networks and Hierarchical Clustering

This paper investigates the optimization of supply chain management using Long Short-Term Memory (LSTM) networks and hierarchical clustering. Focusing on the key issues of demand forecasting and inventory optimization, we initially preprocessed the data, including outlier detection, data merging and encoding, and feature extraction of date data. Subsequently, we constructed an LSTM model to capture long-term dependencies in time series to enhance prediction accuracy. Additionally, we employed hierarchical clustering to categorize products, providing a better understanding of demand characteristics across different product categories. Utilizing the LSTM model, we forecasted demand for the next 15 days and adopted the 1 -wmape metric as the benchmark for prediction accuracy. Experimental results demonstrated that our model achieved an average 1 -wmape of 0.783 across various datasets, validating the model's efficacy. The outcomes of hierarchical clustering revealed differences in demand among product categories, offering valuable insights for inventory management.